Flame Combustion State Detection Of EI-XCL Burner Based On Image Processing

China is a country with large electricity production capacity.In most thermal power plants in China,the main combustion equipment of the furnace is the EI-XCL dual-regulation swirling burner.When the primary air mixed with pulverized coal and fuel enters the furnace and the combustor comes into contact,the combustion directly determines the stability of the entire furnace work.However,at present,we only relies on contact sensors for the combustion state of burners,the data detected by this type of detection method is limited,and only the pulverized coal can be judged whether it is burning,and it cannot provide more real-time data.At present,there are many researches on the flame combustion state detection based on the whole furnace,but there is little research on the combustion state of the EI-XCL burner.Therefore,the EI-XCL burner based on image processing of the combustion state detection study,can further improve the safety and stability of the entire furnace operation.First,this paper describes the research background of image-based flame detection systems for research significance.This paper briely introduced the work environment of the EI-XCL burner and the site profile of the power plant.The feasibility and practical significance of the combustion state detection of the EI-XCL burner based on image are analyzed.Based on this,the problem of how to detect the flame combustion state of the EI-XCL burner is proposed.Based on the problems,the research plan is formulated.Through the work environment and imaging principle of flame television,the gray space is selected as the research basis of this paper.The modified ABDND median filter algorithm was used as the filter of this article to compile and implement the adjustment for the research object of this article.And an impulse noise filter test is performed on the EI-XCL burner flame image in the light-off and low-load(397 MW)operating state.The filtering effect is good,and the filtering effect is excellent in a high noise variance environment,which can provide clear follow-up research flame image.Secondly,this paper proposes an edge detection algorithm based on gray value weights and kirsch operators and compiles them.By performing edge detection experiments on the flame image,and comparing the experimental results with the traditional edge detection algorithm.Comparing the experimental results with the traditional edge detection algorithm,the results show that the proposed method is better.The flame edge threshold obtained by the edge detection algorithm can obtain the area-temperature ratio as the flame image feature value.Then,the average gray value of the flame image,the standard deviation of the flame image gray value,the average foreground transparency of the flame image and Red-blue color component ratio are determined as the five types of flame image samples in this paper.Finally,obtain the eigenvalue samples by extracting the eigenvalues of the flame image samples in this paper.First the K-means algorithm is compiled and implemented.The feature value samples are clustered and classified,and the corresponding classes of each sample in the sample are marked.Then use Libsvm to sample training the sample,and use the classification function to classify the test,the experimental results are more ideal.Therefore,the method of EI-XCL burner flame combustion state detection based on image processing proposed in this paper can meet the combustion state detection of EI-XCL burner under the condition of furnace operation.